MUMBAI, India, Sept. 7 -- Intellectual Property India has published a patent application (202641104301 A) filed by Koneru Lakshmaiah Education Foundation; P. Vijaya Lakshmi; and Dr. Ruth Ramya Kalangi on August 29, 2026, for System And Method For Hyperparameter Optimization Of Bidirectional Lstm Networks For Sentiment Classification.
Inventors include P. Vijaya Lakshmi; and Dr. Ruth Ramya Kalangi.
The application for the patent was published on September 04, 2026, under issue no. 36/2026.
Abstract: ABSTRACT: Title: SYSTEM AND METHOD FOR HYPERPARAMETER OPTIMIZATION OF BIDIRECTIONAL LSTM NETWORKS FOR SENTIMENT CLASSIFICATION A computer-implemented system and method for optimizing sentiment classification of textual data using a Bidirectional Long Short-Term Memory (BiLSTM) neural network is disclosed. The method comprises pre-processing a corpus of text through tokenization, sequence padding and class-weight balancing; constructing a BiLSTM network comprising an embedding layer, a bidirectional recurrent layer, a dropout layer and a dense classification layer; and automatically searching a defined hyperparameter space comprising embedding dimension, number of recurrent units, dropout rate, learning rate and batch size using an automated tuning engine to identify an optimal configuration based on validation accuracy. The optimized network is trained using a weighted loss function and an adaptive optimizer, and evaluated using accuracy, precision, recall and F1-score. The system further generates an interpretability output identifying textual features influencing the predicted sentiment, thereby improving both classification accuracy and explainability. Main Illustrative:
Disclaimer: Curated by HT Syndication.
